[PATCH v3 09/23] iommu/pages: Formalize the freelist API

[Baolu Lu]

On 2/27/25 01:31, Jason Gunthorpe wrote:
> On Wed, Feb 26, 2025 at 02:56:29PM +0800, Baolu Lu wrote:
>> On 2/26/25 03:39, Jason Gunthorpe wrote:
>>> diff --git a/include/linux/iommu.h b/include/linux/iommu.h
>>> index 38c65e92ecd091..e414951c0af83f 100644
>>> --- a/include/linux/iommu.h
>>> +++ b/include/linux/iommu.h
>>> @@ -326,6 +326,18 @@ typedef unsigned int ioasid_t;
>>>    /* Read but do not clear any dirty bits */
>>>    #define IOMMU_DIRTY_NO_CLEAR (1 << 0)
>>> +/*
>>> + * Pages allocated through iommu_alloc_pages_node() can be placed on this list
>>> + * using iommu_pages_list_add(). Note: ONLY pages from iommu_alloc_pages_node()
>>> + * can be used this way!
>>> + */
>>> +struct iommu_pages_list {
>>> +	struct list_head pages;
>>> +};
>>> +
>>> +#define IOMMU_PAGES_LIST_INIT(name) \
>>> +	((struct iommu_pages_list){ .pages = LIST_HEAD_INIT(name.pages) })
>>> +
>>>    #ifdef CONFIG_IOMMU_API
>> Any reason why the above cannot be placed in the iommu-pages.h header
>> file? My understanding is that iommu-pages is only for the iommu drivers
>> and should not be accessible for external subsystems.
> I wanted to do that, but the issue is the gather:
> 
> struct iommu_iotlb_gather {
> 	unsigned long		start;
> 	unsigned long		end;
> 	size_t			pgsize;
> 	struct iommu_pages_list	freelist;
> 
> The struct is inlined so it must be declared. I do not want to include
> iommu-pages.h in this header.
> 
> Once the struct itself is there it made sense to include the INIT too.
> 
> FWIW I have a longstanding desire to split iommu.h into
> internal-driver-facing and external-user-facing files..

Okay, thanks!